Robotics Reference
In-Depth Information
personality to themselves, so it is hardly surprising that early work in this
field indicated that people will respond to robot personalities in the same
way they would respond to similar human personalities and in line with
their preferences in human-human relationships. Also, people find the
experience of interacting with robots that have similar personalities to
their own more satisfying than with robots that have significantly differ-
ent personalities. Simulating and controlling personality is therefore an
essential contribution to the creation of robot companions that will be
acceptable to humans.
Although personality has been the subject of a huge amount of re-
search by psychologists, there is no general agreement amongst them as
to the definition of personality or of its component parts. I quite like
Daniel Rousseau's definition:
Personality characterizes an individual through a set of traits that
influence his or her behaviour. [8]
Traits are persistent characteristics of personality that correspond to gen-
eral patterns of behaviour and modes of thinking, and personality traits
are quite recognizable by observing a person's behaviour, displays of emo-
tion and interpersonal relationships.
Much of the work on simulating personality has been founded partly
on what is known by psychologists as trait theory, in which characteristics
such as sociability and extraversion are seen as determining how we act in
society, and partly on social learning theories, in which our behaviour at
a given moment in time is determined by our situation at that time and
by our past experience in similar situations. Barbara Hayes-Roth and her
group at Stanford University's Knowledge Systems Laboratory have led
much of the research in this field. They developed a personality profiling
model that allows for the specification of traits, such as self-confidence,
gullibility, activity and friendliness, whose intensity can be specified and
varied. This variation in intensity can be specified in terms of numbers,
for example a robot with a gullibility of 100 might be completely and
utterly gullible (“Do you mean I can buy the whole of the Brooklyn
Bridge for $100?”) while one with a gullibility of zero would not believe
anybody (“Surely, just because you're wearing a blue uniform with a shiny
metal badge, a peaked cap and you've got a gun in a holster, you don't
expect me to believe you're a policeman.”)
The traits in the Stanford personality profiling model describe how a
robot feels and how it reacts to other robots and to humans. The traits
Search WWH ::




Custom Search